Human Liver Segments: Role of Cryptic Liver Lobes and Vascular Physiology

www.nature.com/scientificreports

OPEN Human liver segments: role of cryptic liver lobes and vascular physiology in the development of Received: 11 October 2017 Accepted: 16 November 2017 liver veins and left-right asymmetry Published: xx xx xxxx Jill P. J. M. Hikspoors1, Mathijs M. J. P. Peeters1, Nutmethee Kruepunga1,2, Hayelom K. Mekonen1, Greet M. C. Mommen1, S. Eleonore Köhler 1,3 & Wouter H. Lamers 1,4

Couinaud based his well-known subdivision of the liver into (surgical) segments on the branching order of portal veins and the location of hepatic veins. However, both segment boundaries and number remain controversial due to an incomplete understanding of the role of liver lobes and vascular physiology on hepatic venous development. Human embryonic livers (5–10 weeks of development) were visualized with Amira 3D-reconstruction and Cinema 4D-remodeling software. Starting at 5 weeks, the portal and umbilical veins sprouted portal-vein branches that, at 6.5 weeks, had been pruned to 3 main branches in the right hemi-liver, whereas all (>10) persisted in the left hemi-liver. The asymmetric branching pattern of the umbilical vein resembled that of a “distributing” vessel, whereas the more symmetric branching of the portal trunk resembled a “delivering” vessel. At 6 weeks, 3–4 main hepatic-vein outlets drained into the inferior caval vein, of which that draining the caudate lobe formed the intrahepatic portion of the caval vein. More peripherally, 5–6 major tributaries drained both dorsolateral regions and the left and right ventromedial regions, implying a “crypto-lobar” distribution. Lobar boundaries, even in non-lobated human livers, and functional vascular requirements account for the predictable topography and branching pattern of the liver veins, respectively.

Te classical anatomy of the liver is well established, with lef and right lobes separated by the falciform ligament on the diaphragmatic surface, and the round and venous ligaments on the visceral surface of the liver. Te gallbladder, inferior caval vein and liver hilum further delineate the caudate and quadrate lobes. Because these external landmarks were not helpful in planning liver surgery, an anatomical description that was based on the architecture of the biliary and/or vascular trees was developed ~60 years ago1–5. All models use the branching pattern of either the portal and hepatic veins (“French” model1–3) or the bile ducts (“American” model4,5) as segmentation criteria, but difer in their nomenclature. Here we use the accepted “Brisbane 2000” nomenclature6, as modifed by Bismuth (2013). In all concepts, the liver is subdivided into right and lef “hemi-livers” by a plane through the inferior caval and middle hepatic veins, the frst bifurcation of the portal vein or bile duct, and the gallbladder (the Rex-Cantlie “line”)1,3,4,7. Te plane through the inferior caval and right hepatic vein subdivides the right hemi-liver into 2 “sectors”, each being served by 2nd order branches of the portal vein or bile duct. Hjorstjö, however, described the right hemi-liver to contain 3 sectors5, the extra sector arising from the symmetrically distributed 3rd order branches of both bile duct and portal vein in the right anterior sector5,8. Te sectors of the lef hemi-liver difer between the French and American models, and are separated by a plane through the inferior caval vein on the one hand and the lef hepatic vein1 or the umbilical recess/fssure4 on the other (Supplementary Figure 1). All sectors are subdivided into “segments” by a transversely oriented plane through the 1st order branches of the portal vein1 or the distribution of the 3rd order branches of the bile ducts4. Te segments of both models are similar, except that segment 4 is divided into cranial and caudal portions in the American model only4. Couinaud’s

1Department of Anatomy & Embryology, Maastricht University, Maastricht, The Netherlands. 2Department of Anatomy, Faculty of Science, Mahidol University, Rama VI Road, Bangkok, 10400, Thailand. 3NUTRIM Research School of Nutrition and Translational Research in Metabolism, Maastricht, The Netherlands. 4Tytgat Institute for Liver and Intestinal research, Academic Medical Center, Amsterdam, The Netherlands. Correspondence and requests for materials should be addressed to W.H.L. (email: [email protected])

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 1 www.nature.com/scientificreports/

segmental concept of the architecture of the liver1, as translated by Bismuth9,10, with 4 sectors and 8 segments, has become the prevailing standard in textbooks. Nevertheless, the concept remains controversial, because dis- sections and casts show that the “real” boundaries of its segments difer substantially from the “model” bounda- ries11–14. Furthermore, the claim that the right hemi-liver consists of 35 rather than 2 sectors1,3,4 has resurfaced8, while the claim that the lef hemi-liver consists of 3 segments1,2,4 is questioned because the main portal trunk may sprout as many as ~20 direct branches and, therefore, segments14,15. Moreover, the sector boundaries in the lef hemi-liver difer in the French and American models ([1]vs [4]; cf 16;). A newer concept that is based on the distribution of the “main” branches of the portal vein and that acknowledges only 2 main hepatic veins17,18 sidesteps these issues and proposes a 4-lobe segmental model with a caudate lobe, 2 portal segments in right hemi-liver and 1 portal segment in the lef hemi-liver. In all models, the caudate and quadrate lobes stand apart and don’t have well-defned boundaries. Although the discrepancies between the models and the real liver can be solved in a practical manner by imaging of the liver vessels and deducing segmental boundaries from these images19, the conceptual issues appear to arise from a limited understanding of the branching pattern of the liver veins. In line, Majno c.s. and Bismuth recently called for a new anatomical foundation of the segmental architecture of the liver in this journal20,21. Because the main players invoke the embryology of the liver vessels to support their arguments1,2,5,22–24, we rea- soned that a better understanding of the developmental appearance of the liver vessels and architecture would shed light on these issues and allow us to propose a consensus model. In a study of early liver development25, we observed that both vitelline veins become incorporated into the two dorsolateral lobes or wings and both umbilical veins into the single ventromedial lobe of the liver, with the gallbladder marking the midline of the ventromedial lobe. In the 5th week, portal veins started to branch of from the portal stem of the right vitelline vein, from the lef umbilical vein, and from the portal sinus that connects both vessels in the liver hilum. In the present study, we followed the development of these early portal vessels and the appearance of the hepatic veins, and show that the defnitive vascular confguration becomes established in the 7th week. Te branching pattern of the portal and hepatic veins appears to result from the blood-distributing function of the umbilical vein and the (crypto-)lobar architecture of the liver, respectively. We conclude that Couinaud’s model identifes surgically removable quantities of the liver rather than segments that refect the branching order of the portal vein. Results Development of intra-hepatic portal venous system. Right-sided portal tree. The portal vein entered the liver on the right-side of the duodenum (arrowhead in Fig. 1A). From CS14 (34 days of development: Supplementary Fig. 1) onwards, branches from the intrahepatic stem of the portal vein (former right vitelline vein) extended into the liver periphery (blue vessels in Fig. 1), Initially, the number of separate twigs increased (Fig. 1A,C), but with progressing development, 3 branches began to predominate (Fig. 1D): branch #1 extended dorsolaterally, #2 ventrolaterally along the right outer edge of the liver, and branch #3 ventrocranially. Branch #3 was usually a side branch of #2. Te branches supplied the posterior (#1), anterior (#2) and lateral (#3) portions of the right hemi-liver. Comparison of panels A, C and D underlines the continuing expansion of the vessels into the liver periphery with increasing age. Furthermore, higher order branches became apparent (Fig. 1C,D).

Lef-sided portal tree. Te portal twigs that branched of into the lef hemi-liver (peach-colored vessels in Fig. 1B–F and Supplementary Fig. 2) arose from the intrahepatic portion of the lef umbilical vein. Te venous duct, a lef-right shunt, connected the distal end of the umbilical vein in the liver hilum with the inferior caval vein. With time, more and more portal twigs appeared. At CS14, CS15, CS16, and CS18, we counted 4, 6, 6, and 12 such vessels. Between 6.5 (CS18) and 10 weeks, no further increase was observed. Te position of the roots of these portal branches on the main umbilical trunk ascended in a spiraling fashion from anterior to lateral (Fig. 1D). Note that there are no branches on the posterior side of the intrahepatic umbilical vein (Fig. 1F), which refects the establishment of Cantlie’s line (Fig. 1F and Supplementary Figure 2).

Development of hepatic venous system. As the portal veins expanded into the liver periphery, sinusoids in between them began to transform into hepatic veins (compare CS15 (Fig. 1E) with CS16 (Fig. 2A,B)). Tis process spread radially from the suprahepatic portion of the inferior caval vein, with 2 or 3 main hepatovenous trunks emptying into the caval vein (Figs 2–4). Te right hepatic vein expanded caudally into the right dorsolateral side of the liver (Figs 2 and 3). Te middle hepatic vein usually continued as 5 main tributaries: a right medial (ofen called accessory) hepatic vein in the lateral region of the right hemi-liver, a middle hepatic vein with lef and right tributaries, and a lef hepatic vein with lef medial and lef lateral tributaries (Figs 2,3 and 4F, and Supplementary Figure 2). Te topography of the 5 main tributaries of the middle hepatic vein did not difer in the embryos we examined, but they reached the inferior caval vein either separately or afer merging with a neighboring tributary. Two confgurations were commonly observed: the lef medial hepatic branch drained into the lef lateral or the lef tributary of the middle hepatic vein (open black arrowheads in Figs 2 and 3); the corresponding vessel on the right side, the right accessory hepatic vein, drained more proximally or distally into the middle hepatic vein (red arrowhead in Figs 2 and 3). In the 10-week embryo, for example, the middle and lef hepatic veins shared a short common outfow into the inferior caval vein, while the right accessory hepatic vein drained directly into the caval vein (Fig. 3C,D). Te distribution of hepatic veins in the human embryos (Figs 2 and 3) resembled that in adults (Fig. 3E,F), with 2 or 3 hepatic trunks draining directly into the inferior caval vein (Fig. 4), while the main tributaries were symmetrically divided over both hemi-livers (3 branches each; Fig. 4F). Tese fndings show that the hepatovenous drainage pattern in human livers is established at CS17 or CS18.

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 2 www.nature.com/scientificreports/

Figure 1. Te appearance of portal veins in the liver. Right-sided (A), lef-sided (B), ventro-cranial (C–E) and caudal (F) views of CS14 (A,B), CS15 (C,E), and CS18 (D,F) livers. Panel E is C with sinusoids retained. Blue and peach: portal veins in right and lef hemi-livers, respectively. Between CS14 and CS18, the number of large portal branches was pruned to 3 in the right hemi-liver (D) in between the sinusoidal network (E), while >10 branches remained in the lef hemi-liver. Note that portal veins do not cross Cantlie’s line (F). Arrowhead (A): entrance of portal vein into the liver.

Hepatic portion of inferior caval vein. Te sinusoidal network in the developing caudate lobe formed a hepatic vein that joined the inferior caval vein on its posterior side separately from the other main hepatic veins. At CS18 (6.5 weeks) this vessel could be identifed as the intrahepatic portion of the inferior caval vein (Fig. 2 and Supplementary Figure 2).

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 3 www.nature.com/scientificreports/

Figure 2. Hepatic vein development. Anterior (A,C,E) and caudal (B,D,F) views. Formation of 6 main hepatic veins, each identifable by a diferent color, from sinusoidal network between 5.5 (CS16) and 6.5 (CS18) weeks of development. Red and open black arrowheads: variable course of right medial (accessory) and lef medial hepatic veins, respectively; black line (B,D,F): liver contour.

Development of caudate and quadrate lobes. Concurrent with the development of the intrahepatic portion of the caval vein between CS16 and CS18 (5.5–6.5 weeks), the caudate and quadrate lobes became apparent on the posterior side of the liver (Fig. 5). Te caudate lobe developed as a protrusion between the venous duct on its lef side, the intrahepatic portion of the caval vein on its upper right side, and the portal sinus on its caudal side (Fig. 5A). Te portal sinus is a lef-right shunt between the distal end of the intrahepatic umbilical vein and the intrahepatic stem of the portal vein (red arrowhead in Fig. 1D) and will become the transverse part of the portal vein. From CS16 onwards, 2 small portal branches originating from the portal sinus supplied the caudate

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 4 www.nature.com/scientificreports/

Figure 3. Te distribution of the hepatic veins in embryonic and adult human livers. Anterior (A,C) and caudal (B,D–F) views. Panels A–D show the main branches of the hepatic veins in 7 and 10 week-embryos, while panels E,F show the hepatic veins as observed by Fang c.s. (28; reproduced with permission of the publisher). Te color-coded dots in panels E,F identify the hepatic veins in adults that correspond with those in embryonic livers. Each hemi-liver contains 3 hepatovenous branches in both embryos and adults. Color code is identical to that of Fig. 2.

lobe (Fig. 5A–D; Supplementary Figure 2). Concurrent with the transformation of the hepatic vein of the caudate lobe into the intrahepatic portion of the inferior caval vein at CS18, small next-order hepatic veins became the draining vessels of the caudate lobe into the intrahepatic caval vein anteriorly and right laterally (Fig. 5C,E and F). Tere was no defned boundary between the caudate process and the right liver lobe in human livers, but based on the drainage area of hepatic veins26 the caudate process was variable in size among diferent human specimens we studied and could extend in some livers along the entire inferior part of the posterior sector of the right liver lobe (Fig. 5C,E and F).

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 5 www.nature.com/scientificreports/

Figure 4. Various hepatic-vein confgurations draining into the caval vein. Anterior (A–E) views. 3 or 4 hepatovenous trunks and venous duct drain directly into the suprahepatic portion of the inferior caval vein. Te hepatic vein of the caudate lobe (blue) becomes the intrahepatic portion of the caval vein at CS18. Panel F shows the “prototype” of all 6 main hepatic veins. Red interrupted line: diaphragm.

Te quadrate lobe developed as a posterior protrusion from the lef medial region of the liver between the intrahepatic umbilical vein (lef-sided border), gallbladder (right-sided border) and portal sinus (cranial border) (Fig. 5B). It becomes identifable at CS16 or CS17 (cf.27) and is anteriorly continuous with the liver sector between the umbilical fssure laterally and the gallbladder fossa medially. In agreement, the quadrate lobe became served by direct portal branches from the portal sinus at CS18 (Fig. 5D), while the corresponding anterior part is supplied by branches from the umbilical vein (Fig. 6). Blood drained into the lef branch of middle hepatic vein at and afer CS18 (Fig. 5D).

Liver segmentation. Lobated embryonic livers. We compared developing human livers with those of mice because this species has a lobated liver architecture with fssures to defne units (arrows in Fig. 6A). At ED14 (comparable to CS18), mouse livers contained two subdiaphragmatic dorsolateral lobes, a single ventromedial lobe with lef and right halves at each side of the gallbladder, and free caudate and quadrate lobes on the dorsal (posterior) side of the liver (Fig. 6B and Supplementary Figure 2). Te liver hilum was positioned at the intersec- tion between the dorsolateral and ventromedial lobes. Te lef and right dorsolateral lobes were drained by the lef lateral hepatic vein, which was a tributary to the middle hepatic vein, and the right hepatic vein that directly entered the inferior caval vein (Fig. 6D). Te middle hepatic vein drained the ventromedial lobe via a lef medial hepatic tributary, a continuation of the middle hepatic vein with lef and right tributaries and a right medial (accessory) hepatic vein. Te caudate process, an entirely free lobe in the mouse, was drained by the caudate veins that entered the inferior caval vein (Fig. 6B). Te quadrate lobe was poorly developed in the mouse embryo, so that gallbladder and umbilical vein were adjacent structures (Fig. 6A) and Cantlie’s line through the gallbladder and inferior caval vein occupied a more oblique position (Fig. 6C).

Human embryonic liver. In the CS18 human liver, major hepatic veins drained both dorsolateral regions of the liver (Fig. 6F). Te right hepatic vein drained directly into the inferior caval vein, while the lef hepatic vein drained into the middle hepatic vein. Tis middle hepatic vein also drained the ventromedial region of the liver (Fig. 6E) via a right medial (accessory) hepatic vein, a continuation of the middle hepatic vein with lef and right tributaries, and a lef medial hepatic vein. Tis confguration of the main tributaries was observed in all 6 embryos between 6.5 (CS18) and 10 weeks studied, albeit that course of the lateral hepatovenous tributaries of the

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 6 www.nature.com/scientificreports/

Figure 5. Development of the caudate and quadrate lobes. Posterior (A,C,E,F) and posterio-caudal (B,D) views at CS16 (A,B), CS18 (C,D), CS20 (E) and 10 weeks (F). Te caudate lobe (interrupted line) appeared at CS16 and the quadrate lobe (Q; black line) at CS18, were supplied by branches from the portal sinus (black open arrowheads), and were drained by the intrahepatic caval vein and middle hepatic vein (black arrowheads), respectively. Te number and length of hepatic-vein branches that drain unto the intrahepatic caval vein (blue) varies (dotted lines; A,C,E,F) and, therefore, the boundary of the caudate process is variable.

ventromedial region could vary near the inferior caval vein by merging with the tributary from the dorsolateral region. Furthermore, a separate hepatic vein drained the caudate lobe and process (Fig. 5). Tese fndings show that the course of the hepatic veins is, apart from the area directly surrounding the inferior caval vein, comparable in lobated and non-lobated livers, such as the human liver. A similar conclusion can be reached for the portal veins: branch #1 of the right portal vein supplies the right dorsolateral region and branches #2 and #3 the right ventromedial region (Fig. 6C–F). Te numerous portal branches of the umbilical vein branch of in a spiraling

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 7 www.nature.com/scientificreports/

Figure 6. Lobar distribution of veins in human and mouse livers. Cranio-ventral (C,E) and posterior (B,D,F) views. In lobated livers, both caudate lobe and process (green) are free lobes that both drain into the intrahepatic caval vein (B). Te murine quadrate lobe (Q, red) is small (A,B). Te lef and right dorsolateral lobes (B, brown) are drained by the lef lateral and right hepatic veins (D). Te ventromedial lobe (E, red line) is drained by 4 main hepatic branches (lef and right medial, and the two upstream tributaries of the middle hepatic vein). Te distribution of the hepatic veins is similar in mouse and human livers (cf. B–D and E,F). Using the lobar boundaries Couinaud’s surgical segments can be constructed: #1: caudate; #2 and #7: dorsolateral; and ##3–6: ventromedial with 3 hepatic veins. Symbols (A): red arrows: hepatic fssures; IVC: intrahepatic caval vein, UV: umbilical vein, green: gallbladder. Color code veins is as in Figs 1 and 2.

fashion, with a generally anterior direction when passing the ventromedial region of the liver to a lateral direction when close to the dorsolateral region (Fig. 6C–F). Importantly, we did not observe the anticipated hierarchical binary division of the portal veins that underlies the segmental concept of the liver architecture.

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 8 www.nature.com/scientificreports/

Discussion Branches of the portal vein became identifable at CS14 (~34 days), while the hepatic veins formed 5 days later (CS16 or ~39 days). Te branching pattern of the portal veins in the right and lef hemi-livers was initially similar. However, afer the appearance of the hepatic veins and, hence, presumably afer the development of a higher blood fow rate through the liver periphery, three main portal branches emerged in the right hemi-liver, whereas the original branching pattern, with >10 portal veins branching from the intrahepatic portion of the umbilical vein, was preserved in the lef hemi-liver. Tis confguration can also be recognized in the drawing of the liver vessels of a CS19 embryo by Lassau23 and is identical to the confguration described by Fasel for adult human livers15. Similarly, by CS17 or CS18 the hepatic veins had established a confguration that was identical to that shown earlier for adult liver28,29. Te venous systems of the liver have, therefore, acquired the adult confguration at 6.5 weeks of development (CS18).

The branching pattern of the portal veins in the right and left hemi-liver. As a consequence of the appearance of the large intrahepatic lef-right shunts and hepatic veins, the aferent blood supply to the liver via the intrahepatic portion of the umbilical vein came to lie upstream of the very limited blood supply via the portal vein (at 8 weeks, the diameter of the extrahepatic portion of the portal vein amounted to only 10–15% of that of the umbilical vein). Te branching pattern of the umbilical vein was highly asymmetric, with a large continuing stem and relatively small portal branches forking of serially (“monopodial” branching pattern), whereas the branching pattern of the portal vein in the right hemi-liver was more symmetric. Te highly asymmetric branching pattern of the intrahepatic portion of the umbilical vein, with large branching angles and small diameters of the portal branches, resembled that of a transport (“distributing”) vessel, whereas the more symmetric branching of the portal vein in the right hemi-liver better resembled that of tissue-supplying (“delivering”) vessels30. A “distributing” confguration appears to minimize pumping power (power-cost optimization) and seize of the vascular tree (material-cost optimization)31. A large diference in diameter and length of daughter branches in non-pulsatile (venous) fow systems increases the efciency of the system during growth (decreases the allometric constant)32. Although these fndings clearly point at the importance of hemodynamics for liver development and vascular architecture, no fow measurements are available until the fetus is 19 weeks old33. Te observed monopodial branching pattern of the umbilical vein is, nevertheless, clearly at odds with the supposedly dichotomous branching pattern of the portal vein that underlies Couinaud’s segmental liver anatomy.

The branching pattern of the hepatic veins. Like most studies34, we observed 2 or 3 main hepatic-vein outlets into the inferior caval vein. If only 2 stems were present, the third vein merged with the middle hepatic vein near its exit from the liver. More peripherally, we and others8,35 observed a more constant distribution of the hepatic veins, with 5 or 6 major tributaries draining Couinaud’s segments ##2–7. Te “extra” peripheral hepatovenous tributaries also merged with the middle or lef hepatic vein near their exit into the caval vein. We hypothesize that the more pronounced variability in the topography of the hepatic veins near their exit into the caval vein compared to that in the liver periphery relates to their association in the periphery with the dorsolateral and ventromedial liver regions that we described in early liver development25 and that are separated by fssures in lobated livers24. While identifable veins drain the dorsolateral regions, the ~4 hepatic veins that drain the large ventromedial region or lobe of the liver appear to follow a monopodial branching pattern, with the lef medial and right “accessory” veins as frst branches and the main axis extending anteriorly to split in lef and right branches on top of the gallbladder (Fig. 4F). Based on the rather predictable branching pattern of the hepatic veins that resembles that in lobated livers, we posit that the human liver is “crypto-lobated”. Apparently, the lobar “guidance” for the course of the hepatic veins is not present in the immediate surrounding of the caval vein. Te main hepatic vein of the caudate lobe transformed into the intrahepatic portion of the inferior caval vein. Tis fnding explains why ofen several relatively small hepatic veins from the caudate lobe drain directly into the caval vein36. It further demonstrates that the intrahepatic portion of the inferior caval vein derives from the most posterior hepatic vein. Together with the origin of the suprahepatic portion of the inferior caval vein from the right hepatocardiac channel, which is the common connection of the vitelline and umbilical veins with the venous sinus of the heart25, these fndings defnitively establish the origin of the inferior caval vein between the well-established right-sided subcardinal vein near the kidney and its entrance into the right atrium.

The relation of liver lobes to Couinaud’s hepatic segments. Te human liver is not lobated24. Te only lobes that are identifable are the caudate and, to a lesser extent, the quadrate lobes. However, malformed livers with identifable lobes are sometimes seen37. In lobated livers, the hepatic veins are found in the center of each lobe38, whereas they defne the peripheral boundaries of the surgical segments of the liver. Tis discrepancy reveals the diference between the lobar and segmental anatomy of the liver. Te lobar territories can, nevertheless, be identifed in the non-lobated human liver. Both dorsolateral lobes (segments #2 and #7 in Fig. 6) can be recognized in adult livers by the topography of the dorsolateral branches of the portal vein (e.g. Figures 1,2 in12). On the lef side, this vessel is the most distal branch of the umbilical portion of the lef portal vein. In this respect, Couinaud’s “unnatural” lef portal fssure39 properly defnes this boundary. Te hepatic vein draining the lef dorsolateral lobe is the lef lateral hepatic vein. Te situation in the right hemi-liver is less clear-cut, but segment #7 probably represents the right dorsolateral lobe40. Tere is, however, no landmark vessel to delimit this segment from the remaining part of Couinaud’s posterior sector. For this reason we distinguished only 7 segments in Fig. 6. Similarly, the Spigelian sub-lobe of the caudate lobe has a distinct lef boundary, but the rightward boundary of the caudate process is not so apparent. If we defne the caudate process by its drainage territory (Fig. 5; cf.26), it can be much larger than usually shown, does resemble Couinaud’s segment #6, and is similar to the caudate process of lobated livers (Fig. 6). In this respect it is of interest that segment #6 develops, according to Couinaud, relatively late and is most prominent in human livers39.

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 9 www.nature.com/scientificreports/

Couinaud’s segments ##3–5, and #8 correspond with the ventromedial lobe of lobated livers. Te gallbladder and middle hepatic vein demarcate Cantlie’s line. Since no portal vein branches passed this line (Figs 1 and 6), it is an accepted plane to separate the ventromedial lobe into lef and right halves. However, no good landmarks are available to further divide the lef and right halves of this large lobe into sectors. In the American model, the umbilical vein or round ligament serves as a sector boundary to subdivide the lef-sided portion of the ventromedial lobe. Because part of the umbilical vein persists as the umbilical portion of the lef portal vein, it forced Couinaud to come up with a very small lef lateral hepatic sector (segment #216;). However, the umbilical portion of the lef portal vein is special as it lies very superfcial and can be dissected free16. It is, therefore, probably not suitable as a landmark to subdivide the lef-sided part of the ventromedial lobe into segments #3 and #4 would, if it were only, because this criterion would qualitatively difer from that (the right medial (accessory) hepatic vein) used to subdivide the right-sided part of the ventromedial lobe into segments #5 and #6. As we argued above, however, these hepatic veins follow a rather predictable course in the ventromedial lobe or region, so that we can identify 4 sectors. Using the portal and hepatic veins as criteria to delineate liver regions, we can, therefore, only identify the 7 sectors described by Hjortsjö5,8. In humans, the quadrate lobe develops between the gallbladder and umbilical vein27, but this lobe remains a rudimentary structure in mice (Fig. 6A,B), so that the umbilical vein (round ligament afer birth) lies close to the gallbladder. Such a confguration is also seen in humans with a rudimentary quadrate lobe27.

Is there a developmental origin of the surgical liver segments?. Te discrepancy between the branching pattern of the portal vein and the distribution of hepatic segments contradicts the putative embryological origin of Couinaud’s segmental liver anatomy39. Instead, the striking similarity of hepatic segments based on the distribution of the portal and hepatic veins (the French model) or on the distribution of the bile ducts and hepatic arteries (the American model) is truly remarkable because of the large time window between the appearance of the liver veins and the bile ducts/arteries: the intrahepatic bile ducts start to remodel into tubular structures only afer 10 weeks of development and extend only slowly into the liver periphery41–44. Te development of the hepatic artery accompanies bile duct development41. Te similarity of the liver segments according to the French and American models suggests, instead, that they share a functional basis, e.g. that imposed by perfusion and metabolic demands. Accordingly, a comparison of Fasel’s and our models shows that the initially similar-sized portal veins in the lef hemi-liver increase diferentially in size, with some becoming much larger than others. We, therefore, suggest that Couinaud’s segments represent similar sized liver units (8–15% of the total liver volume, with only segment #8 being much larger (24%45;) that are supplied by hepatic arteries and an adaptively enlarged subpopulation of portal vein branches. We further posit that liver segments as defned in the French and American models represent surgically removable quantities of tissue with fractal properties: a large surgical segment (e.g. #8) can be split in 2 subsegments with the same approach as used to delineate the regular surgical segments46. Conclusions Te venous systems of the liver acquire the adult confguration at 6.5 weeks of development. Te topography of the portal and hepatic veins was similar to that in lobated livers, showing that the human liver is crypto-lobated. Te monopodial branching pattern of the umbilical vein is not compatible with the dichotomous branching pattern of the portal vein that underlies Couinaud’s segmental liver anatomy. We, therefore, concluded that Couinaud’s model does not have an embryological origin. Te similarity of the segmental boundaries of the French and American models suggests, instead, that an adaptable mechanism underlies the segmental anatomy of the liver, e.g. that imposed by perfusion and metabolic demands. Materials and Methods Ethics. Tis study was undertaken in accordance with the Dutch regulations for the proper use of human and animal tissue for medical research purposes. Anonymized specimens were included from the historical collections of human embryos of the Departments of Anatomy and Embryology, Leiden University Medical Center (LUMC), Leiden, and the Academic Medical Center (AMC), Amsterdam, Te Netherlands, that were donated for scientifc research. Te Dutch collections that we cite as sources were established in the 1950s–1970s by the late professors Dankmeijer (Leiden) and Los (Amsterdam). Furthermore, we have included a historical collection of mouse embryos from the Department of Anatomy and Embryology, AMC, Amsterdam, which was established in the 1970s–1980s by the late professor Los (Amsterdam). Te study of historical collections is exempt from approval by a Medical Ethics Committee in the Netherlands. In addition, human embryos from the Carnegie Collection, Washington D.C., USA were included from the Digitally Reproduced Embryonic Morphology (DREM) project (Dr John Cork; Cell Biology & Anatomy, LSU Health Sciences Center, New Orleans; https:// www.ehd.org/virtual-human-embryo/about.php, http://virtualhumanembryo.lsuhsc.edu), who made digitized sections available to us.

Embryos. We reconstructed 12 human embryos between 5 and 10 weeks of development that represented high-quality specimens of the historical collections of AMC and LUMC, which together contain over 150 human embryos. In addition, the digitized human embryos of the Carnegie Collection were used. Te criteria of O’Rahilly as modifed in 2010 were used to determine the Carnegie stage of development47. Brain development48 and the return of the physiological hernia between 9 and 9.5 weeks of development49 were used to estimate the age of the embryos afer CS23. Specimens of 8–10 weeks development are also referred to as embryos. Because most mammalian species difer from human in having lobated livers with fssures that can serve as topographic landmarks, we also studied mouse embryos from the AMC collection.

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 10 www.nature.com/scientificreports/

Image acquisition, 3D-reconstruction and visualization. Digitization of the stained sections, pro- cessing of the digital images, and voxel size calculation were performed as described50. AMIRA (version 6.2; base package; FEI Visualization Sciences Group Europe, Merignac Cedex, France) was used to generate 3D reconstructions. Delineation of the liver contour, blood vessels/networks, and structures of interest was performed manually (Supplementary Figure 1). Polygon meshes from all reconstructed materials of an embryo were smoothened (Supplementary Figure 1) and exported via ‘vrml export’ to Cinema 4D (MAXON Computer GmbH, Friedrichsdorf, Germany). Te accuracy of the remodeling process was safeguarded by simultaneous visualization in Cinema 4D of the original output from Amira and the remodeled Cinema model (Supplementary Figure 1). Subsequently, the Cinema 3D-model was exported via ‘wrl export’ to Adobe portable device format (PDF) reader version 9 (http://www.adobe.com) for the generation of 3D-interactive PDF fles (Supplementary Figure 2). References 1. Couinaud, C. (Masson, Paris, 1957). 2. Gans, H. Introduction to hepatic surgery. (Elsevier, 1955). 3. Goldsmith, N. A. & Woodburne, R. T. Te surgical anatomy pertaining to liver resection. Surg Gynecol Obstet 105, 310–318 (1957). 4. Healey, J. E. Jr. & Schroy, P. C. Anatomy of the biliary ducts within the human liver; analysis of the prevailing pattern of branchings and the major variations of the biliary ducts. AMA Arch Surg 66, 599–616 (1953). 5. Hjortsjö, C. H. Te topography of the intrahepatic duct systems. Acta Anat 11, 599–615 (1951). 6. Terminology Committee of the IHPBA & Te Brisbane 2000 terminology of liver anatomy and resections. Surg 2, 333–339 (2000). 7. Cantlie, J. On a new arrangement of the right and lef lobes of the liver. Proceedings of the Anatomical Society of Great Britian and Ireland 32, 4–9 (1897). 8. Kogure, K., Kuwano, H., Fujimaki, N., Ishikawa, H. & Takada, K. Reproposal for Hjortsjo’s segmental anatomy on the anterior segment in human liver. Arch Surg 137, 1118–1124 (2002). 9. Bismuth, H. Surgical anatomy and anatomical surgery of the liver. World J Surg 6, 3–9 (1982). 10. Bismuth, H. Revisiting liver anatomy and terminology of hepatectomies. Ann Surg 257, 383–386, https://doi.org/10.1097/ SLA.0b013e31827f171f (2013). 11. Fasel, J. H. et al. Segmental anatomy of the liver: poor correlation with CT. Radiology 206, 151–156, https://doi.org/10.1148/ radiology.206.1.9423665 (1998). 12. Platzer, W. & Maurer, H. Zur segmenteinteilung der leber. Acta Anat 63, 8–31 (1966). 13. van Leeuwen, M. S. et al. Portal venous and segmental anatomy of the right hemiliver: observations based on three-dimensional spiral CT renderings. Am J Roentgenol 163, 1395–1404, https://doi.org/10.2214/ajr.163.6.7992736 (1994). 14. Fasel, J. H. & Schenk, A. Concepts for liver segment classifcation: neither old ones nor new ones, but a comprehensive one. J Clin Imaging Sci 3, 48, https://doi.org/10.4103/2156-7514.120803 (2013). 15. Fasel, J. H. Portal venous territories within the human liver: an anatomical reappraisal. Anat Rec 291, 636–642, https://doi. org/10.1002/ar.20658 (2008). 16. Strasberg, S. M. Terminology of liver anatomy and liver resections: coming to grips with hepatic Babel. J Am Coll Surgeons 184, 413–434 (1997). 17. Takasaki, K. Glissonean pedicle transection method for hepatic resection: a new concept of liver segmentation. J Hepato-Biliary-Pan 5, 286–291 (1998). 18. Cho, A. et al. Proposal for a reclassifcation of liver based anatomy on portal ramifcations. Am J Surg 189, 195–199, https://doi. org/10.1016/j.amjsurg.2004.04.014 (2005). 19. Soler, L., Mutter, D., Pessaux, P. & Marescaux, J. Patient specifc anatomy: the new area of anatomy based on computer science illustrated on liver. J Vis Surg 1, 1–12 (2015). 20. Bismuth, H. A new look on liver anatomy: needs and means to go beyond the Couinaud scheme. J Hepatol 60, 480–481, https://doi. org/10.1016/j.jhep.2013.12.010 (2014). 21. Majno, P. et al. Anatomy of the liver: an outline with three levels of complexity–a further step towards tailored territorial liver resections. J Hepatol 60, 654–662, https://doi.org/10.1016/j.jhep.2013.10.026 (2014). 22. Elias, H. & Petty, D. Gross anatomy of the blood vessels and ducts within the human liver. Am J Anat 90, 59–111, https://doi. org/10.1002/aja.1000900104 (1952). 23. Lassau, J. P. & Bastian, D. Organogenesis of the venous structures of the human liver: a hemodynamic theory. anatomia clinica 5, 97–102 (1983). 24. Nettelblad, S. C. Die lobierung und innere topographie der saugerleber. Acta Anat 21(suppl 20), 7–246 (1954). 25. Hikspoors, J. P. J. M. et al. Te fate of the vitelline and umbilical veins during the development of the human liver. J Anat, n/a-n/a, https://doi.org/10.1111/joa.12671 (2017). 26. Benko, T. et al. Portal supply and venous drainage of the caudate lobe in the healthy human liver: virtual three-dimensional computed tomography volume study. World J Surg 41, 817–824, https://doi.org/10.1007/s00268-016-3791-8 (2017). 27. Tompson, P. Te development of the lobus quadratus of the liver, with special reference to an unusual anomaly of this lobe in the adult. J Anat Physiol 48, 222–237 (1914). 28. Fang, C. H. et al. Anatomical variations of hepatic veins: three-dimensional computed tomography scans of 200 subjects. World J Surg 36, 120–124, https://doi.org/10.1007/s00268-011-1297-y (2012). 29. Lhuaire, M. et al. Developmental anatomy of the liver from computerized three-dimensional reconstructions of four human embryos (from Carnegie stage 14 to 23). Ann Anat 200, 105–113, https://doi.org/10.1016/j.aanat.2015.02.012 (2015). 30. Zamir, M. Distributing and delivering vessels of the human heart. J Gen Physiol 91, 725–735 (1988). 31. Tekin, E., Hunt, D., Newberry, M. G. & Savage, V. M. Do vascular networks branch optimally or randomly across spatial scales? PLoS Comput Biol 12, e1005223, https://doi.org/10.1371/journal.pcbi.1005223 (2016). 32. Brummer, A. B., Savage, V. M. & Enquist, B. J. A general model for metabolic scaling in self-similar asymmetric networks. PLoS Comput Biol 13, e1005394, https://doi.org/10.1371/journal.pcbi.1005394 (2017). 33. Kessler, J., Rasmussen, S., Godfrey, K., Hanson, M. & Kiserud, T. Venous liver blood fow and regulation of human fetal growth: evidence from macrosomic fetuses. Am J Obstet Gynecol 204(429), e421–427, https://doi.org/10.1016/j.ajog.2010.12.038 (2011). 34. Germain, T. et al. Liver segmentation: practical tips. Diagn Interv Imaging 95, 1003–1016, https://doi.org/10.1016/j.diii.2013.11.004 (2014). 35. Shindoh, J. et al. Vascular architecture in anomalous right-sided ligamentum teres: three-dimensional analyses in 35 patients. HPB (Oxford) 14, 32–41, https://doi.org/10.1111/j.1477-2574.2011.00398.x (2012). 36. Kogure, K., Kuwano, H., Fujimaki, N. & Makuuchi, M. Relation among portal segmentation, proper hepatic vein, and external notch of the caudate lobe in the human liver. Ann Surg 231, 223–228 (2000). 37. Bartsch, F., Ackermann, M., Lang, H. & Heinrich, S. Unfused liver segments: a case report of an unknown phenotype of the Conradi- Hunermann-Happle Syndrome. J Gastrointestin Liver Dis 25, 547–549, https://doi.org/10.15403/jgld.2014.1121.254.sch (2016).

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 11 www.nature.com/scientificreports/

38. Oishi, Y. et al. Anatomical evaluation of hepatic vascular system in healthy beagles using X-ray contrast computed tomography. J Vet Med Sci 77, 925–929, https://doi.org/10.1292/jvms.14-0469 (2015). 39. Couinaud, C. Liver anatomy: portal (and suprahepatic) or biliary segmentation. Dig Surg 16, 459–467 (1999). 40. Kogure, K., Ishizaki, M., Nemoto, M., Kuwano, H. & Makuuchi, M. A comparative study of the anatomy of rat and human livers. J Hepatobiliary Pancreat Surg 6, 171–175 (1999). 41. Gouysse, G. et al. Relationship between vascular development and vascular diferentiation during liver organogenesis in humans. J Hepatol 37, 730–740 (2002). 42. Nakanuma, Y., Hoso, M., Sanzen, T. & Sasaki, M. Microstructure and development of the normal and pathologic biliary tract in humans, including blood supply. Microsc Res Tech 38, 552–570, https://doi.org/10.1002/(SICI)1097-0029 (19970915)38:6<552::AID- JEMT2>3.0.CO;2-H (1997). 43. Roskams, T. & Desmet, V. Embryology of extra- and intrahepatic bile ducts, the ductal plate. Anat Rec (Hoboken) 291, 628–635, https://doi.org/10.1002/ar.20710 (2008). 44. Costa, A. M., Pegado, C. S. & Porto, L. C. Quantifcation of the intrahepatic biliary tree during human fetal development. Anat Rec 251, 297–302 (1998). 45. Tani, K. et al. Venous drainage map of the liver for complex hepatobiliary surgery and liver transplantation. HPB (Oxford) 18, 1031–1038, https://doi.org/10.1016/j.hpb.2016.08.007 (2016). 46. Lee, J. H., Jin, G. Y., Jin, Z. W., Yu, H. C. & Cho, B. H. Ramifcation of Glisson’s sheath peripheral branches and clinical implications in the era of local ablation therapy. Surg Radiol Anat 32, 911–917, https://doi.org/10.1007/s00276-010-0643-3 (2010). 47. O’Rahilly, R. & Muller, F. Developmental stages in human embryos: revised and new measurements. Cells Tissues Organs 192, 73–84, https://doi.org/10.1159/000289817 (2010). 48. Bayer, S. A. & Altman, J. Te Human Brain During the Early First Trimester. (CRC Press, 2007). 49. Sofers, J. H., Hikspoors, J. P., Mekonen, H. K., Koehler, S. E. & Lamers, W. H. Te growth pattern of the human intestine and its mesentery. BMC Dev Biol 15, 31, https://doi.org/10.1186/s12861-015-0081-x (2015). 50. Hikspoors, J. P. et al. Development of the human infrahepatic inferior caval and azygos venous systems. J Anat 226, 113–125, https:// doi.org/10.1111/joa.12266 (2015). Acknowledgements We thank Drs Maurice van den Hof (AMC), Marco de Ruiter (LUMC) and John Cork (DREM project) for allowing us to use their institutional series of human embryos. Special thank goes to Els Terwindt (MU) for her technical assistance. Te fnancial support of ‘Stichting Rijp’ is gratefully acknowledged. Author Contributions J.H. participated in collection, analysis and visualization of data, and wrote the manuscript. M.P. has made substantial contributions to the collection and visualization of data. N.K. and H.M. participated in analysis and interpretation of data. G.M. and J.H. were responsible for formatting all fgures. S.E.K. participated in analysis of data, provided guidance, assisted with data interpretation and edited the manuscript. W.L. conceived the study, provided guidance and assisted with data interpretation, and helped write and edit the manuscript. Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-017-16840-1. Competing Interests: Te authors declare that they have no competing interests. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. Te images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

SCIENTIfIC REPOrtS | 7:17109 | DOI:10.1038/s41598-017-16840-1 12